Data Fabric
Alan McSweeney
http://ie.linkedin.com/in/alanmcsweeney
What Is An Enterprise Data Fabric?
18 February 2018 3
Core Data Fabric Conceptual Model
18 February 2018 4
Data Fabric Conceptual Model – Components - 1 of
2
Component Description
External Interacting Parties These are the range of external parties that supply data to and access data from the enterprise
External Party Interaction These are the set of applications and data interface and exchange points provided specifically to
Zones, Applications, Channels External Interacting Parties to allow them supply data to and access data from the enterprise
and Facilities
These can be hosted internally or externally or a mix of both
External Third Party These are third-party applications (such as social media platforms) that contain information
Applications about the enterprise or that are used by the enterprise to present information to or interact with
External Interacting Parties or where the enterprise is referred to, affecting the perception or
brand of the enterprise
External Data Sensors Sources of remote data measurements
External Party Interaction Zones These are applications and sets of data created by the enterprise to be externally facing where
Data Stores external parties can access information and interact with the enterprise
External Devices These are devices connected with services offered by the enterprise (such as ATMs and Kiosks)
Date Intake/Gateway This is the set of facilities for handling data supplied to the enterprise including validation and
transformation including a possible integration or service bus
18 February 2018 5
Data Fabric Conceptual Model – Components - 2 of
2
Component Description
Line of Business Applications This represents the set of line of business applications deployed on external infrastructure used
Hosted Outside the Organisation by business functions to operate their business processes This includes cloud facilities such as
external data storage and XaaS facilities and an integration service to connect external data to
internal data
External Application Operational These are the various operational data stores used by the Line of Business Applications used by
Data Stores Line of Business Applications Hosted Outside the Organisation
Data Mastering These are facilities to create and manage master data and data extracted from operational data
to create a data warehouse and data extracts for reporting and analysis. This includes an extract,
transformation and load facility
18 February 2018 7
Why Create A Conceptual Data Fabric Model?
• Conceptual data fabric model represents a rich picture of the enterprise’s data
context
− Embodies an idealised and target data view
• Detailed visualisations represent information more effectively than lengthy
narrative text
− More easily understood and engaged with
• Show relationships, interactions
• Capture complexity easily
• Provides a more concise illustration of state
• Better tool to elicit information
• Gaps, errors and omissions more easily identified
• Assists informed discussions
• Evolve and refine rich picture representations of as-in and to-be situations
• Cannot expect to capture every piece of information – focus on the important
elements
• A rich picture is not a data management process map (yet)
18 February 2018 8
Differences Between Current And Target Conceptual
Data Model
• Use the conceptual data fabric model to identify gaps
between the current and desired target
18 February 2018 9
Core Data Fabric Conceptual Model
18 February 2018 10
Responding To Interrelated Data Trends
Internal and
External Digital
Expectations,
Analytics
Capabilities
18 February 2018 11
Responding To Interrelated Data Trends
18 February 2018 12
IT Function Data Leadership
18 February 2018 13
What Are The Data Challenges?
18 February 2018 14
Data Fabric Is Much More Than A Move To The
Cloud
• Enterprise data fabric should enables appropriate and seamless
move to multiple cloud/XaaS platforms - public, private and
hybrid - across the entire data infrastructure
− Storage
− Business applications
− Data management
− Reporting and analytics tools
• Cloud impacts the enterprise’s approach to data
− Enterprises cannot ignore cloud and XaaS options
• Enterprise data fabric needs to encompass the diversity of data
storage infrastructures
• Design an open and flexible data fabric that improves the
responsiveness of the IT function and reduces shadow IT
18 February 2018 15
Why Have An Enterprise Data Fabric?
18 February 2018 17
Creating A Data Vision
18 February 2018 18
Creating A Data Vision – Making The Enterprise Data
Focussed
• Enable value to be derived from data
− Shorten the distance between business and analytics
• Facilitate data initiatives by removing the barriers to data
enablement
• IT needs to understand the data needs and associated data
business processes of the business and deliver results
− IT showing data leadership
• Top-down visualisation that is then implemented by
appropriate components are different layers
18 February 2018 19
Current Data Fabric State
18 February 2018 20
Target Data Fabric Future State
18 February 2018 21
Achieving The Target Data Fabric State
18 February 2018 23
Key Design Principles Of A Data Fabric
Have A Choice Of
React and Move And Be Able To
Quickly Adopt New
Technologies
Business IT
Offer Innovative
React and Move
Facilities and
Substantially
Functions
React Quickly To
Enable Growth
New
Opportunities
Requirements
18 February 2018 25
Data Fabric Is A Basic Building Block Of An Enterprise
Data Strategy
}
Insight/
Reporting
Forecast
You Cannot
Have This ...
Monitoring Analysis
... Without
Data Architecture This
Management
Data Governance
Data Warehousing and Business Intelligence
Management Solid
Data Security Management Data
Reference and Master Data Management
Management
Foundation
Document and Content Management and
Framework
Metadata Management
Data Development
Insight/
Reporting
Forecast
Monitoring Analysis
What Is Currently
Happening? Why It Happened?
Reporting
• Report on Gathered Information On What Happened
To Understand Pinch Points, Quantify Effectiveness,
Measure Resource Usage And Success
Monitoring Trailing
• Gather Information In Realtime To Understand Indicators
Activities, Respond And Make Reallocation Decisions
Analysis
• Understand Reasons For Outcomes and Modify
Operation To Embed Improvements Leading
Indicators
Insight and Forecast
• Quantify Propensities, Forecast Likely Outcomes,
Identify Leading Indicators, Create Actionable
Intelligence
February 18, 2018 28
Objective Of Designing An Enterprise Data Fabric
18 February 2018 29
Extended Data Fabric Conceptual Model
Administration,
Management
Monitoring,
Logging Alerting, Event
Management
Archival,
Recall
18 February 2018 30
Extended Data Fabric Conceptual Model
18 February 2018 32
Data Lifecycle View
18 February 2018 34
Data As A Realisable Asset
• Raw data must be refined into a format that can be used in order to
be viewed as an asset with realisable value
• For data to be an asset it must be:
− Have its underlying value extracted
− Accessible
− Usable
• Data has physical and tangible characteristics:
− Mass – it has bulk and requires resources to store, process and move
− Heat – it gets cold over time with different levels of dissipation
− Energy – data has different levels of energy based on its movement and value
− Volatility – the underlying value of the data can be lost at differing rates
− Complexity – the content and structure of the data is variable
− Motion – data moves from location to location as it is generated, stored,
process
− Structure – data may be structured, semi-structured or high-structured
− Size to Value Ratio – the usable value with the data may be large or small
relative to the volume of the raw data
18 February 2018 35
External Interacting Parties
18 February 2018 36
External Interacting Parties
18 February 2018 37
External Party Interaction Zones, Applications,
Channels and Facilities
18 February 2018 38
External Party Interaction Zones, Applications,
Channels and Facilities
• This is the range of application-based modes and methods
of interaction between the enterprise and the External
Interacting Parties (rather than pure email)
18 February 2018 39
External Party Interaction Zones Data Stores
18 February 2018 40
External Party Interaction Zones Data Stores
18 February 2018 41
Date Intake/Gateway
18 February 2018 42
Date Intake/Gateway
18 February 2018 43
External Third Party Applications
18 February 2018 44
External Third Party Applications
18 February 2018 45
External Data Sensors
18 February 2018 46
External Data Sensors
18 February 2018 47
External Devices
18 February 2018 48
External Devices
18 February 2018 49
Line of Business Applications
18 February 2018 50
Line of Business Applications
18 February 2018 51
Data Storage Platforms
18 February 2018 52
Data Storage Platforms
18 February 2018 53
Data Reporting and Analysis Facilities
18 February 2018 54
Data Reporting and Analysis Facilities
18 February 2018 55
Document Management Systems And Document
Sharing and Collaboration
18 February 2018 56
Document Management Systems And Document
Sharing and Collaboration
• This represents the facilities to store structure and
unstructured document-oriented data including document
metadata, extract information from documents and
support ad hoc and formal workflows related to these
documents
18 February 2018 57
Desktop Applications
18 February 2018 58
Desktop Applications
18 February 2018 59
Many Data Types
18 February 2018 60
Data Fabric As Data Plumbing And A Data Refinery
18 February 2018 61
Data Layers Across Data Fabric
Layer Components Data Scope
Layer 8+ Data Operations, Usage, Unified management across all environments and all
Management, Control, layers and ensure performance, availability,
Governance, Analysis, Modelling reliability, scalability, maintainability and
supportability
Layer 7 Data Presentation, Platforms, Set of data accessing and data using business
Applications, Systems and Business applications
Processes
Layer 6 Data Security and Governance Implement common data security policies across all
environments and platforms
Layer 5 Data Logical Access and Integration Insulate and abstract access from knowledge of
environments and platforms and integrate data
systems and data management
Layer 4 Data Transportation Provide a common data transport that connects all
environments
Layer 3 Data Network and Connectivity Connections to storage and physical access
irrespective of location across entire network
Layer 2 Data Physical Access Provide physical access to data on storage layer
Layer 1 Data Storage and Transmission Store data transparently on multiple environments
Infrastructure and move data between environments
18 February 2018 62
Building A Comprehensive Data Vision
Strategy Area
Strategy Area
Component
Component Type …
Component
Core Data Fabric Conceptual
Model Components …
Component
Component Type …
Data Management and
Comprehensive Data Vision Component
Operations Facility
Extended Data Fabric
Conceptual Model …
Data Management and
Operations Facility
Stage
Data Lifecycle …
Stage
Type
Data Types …
Type
18 February 2018 63
Extending Conceptual Model To Additional Levels Of
Detail To Build A Comprehensive Data Vision
• Individual data views can be combined to articulate a
comprehensive data vision
− Enterprise Data Strategy
• Individual strategy areas
− Core Data Fabric Conceptual Model Components
• Individual elements within each component
− Extended Data Fabric Conceptual Model
• Operating principles and interactions
− Data Lifecycle
• Individual stages within lifecycle
− Data Types
• Individual data types
• Builds an understanding of how the enterprise wants and
needs to handle and use data
18 February 2018 64
Extending Conceptual Model To Additional Levels Of
Detail To Build A Comprehensive Data Vision
Additional
Data
Dimensions
and Views
18 February 2018 65
Summary
• Data fabric is concerned with creating a data vision for the enterprise
• The conceptual data fabric model represents a rich picture of the enterprise’s data
context
− Detailed visualisations represent information more effectively than lengthy narrative text
• Use the conceptual data fabric model to identify gaps between the current and
desired target
• Data fabric provides a basis for understanding the enterprise’s ideal data
architecture
• Designing a data fabric enables the enterprise respond to and take advantage of
key related data trends
− Shadow IT occurs when the IT function cannot deliver IT change and new data facilities
quickly
− Uncontrolled data platforms and storage represent a significant and real risk to the
enterprise
• Enterprise data fabric should enables appropriate and seamless move to multiple
cloud/XaaS platforms - public, private and hybrid - across the entire data
infrastructure
• Enables the enterprise focus on achieving benefits from data rather than on data
operations
18 February 2018 66
More Information
Alan McSweeney
http://ie.linkedin.com/in/alanmcsweeney
18 February 2018 67